基于背景知识的全球长时间序列反照率反演
作者简介:商 荣(1990-),男,湖北武汉人,博士生,研究方向为定量遥感。E-mail: shangr.13b@igsnrr.ac.cn
收稿日期: 2015-03-06
要求修回日期: 2015-05-27
网络出版日期: 2015-11-10
基金资助
气象行业科研专项(GYHY201106014)
中国科学院战略性先导科技专项(XDA05090303)
国家自然科学基金项目(41171285、41301354)
Generation of Global Long-term Albedo Product Based on the Background Knowledge
Received date: 2015-03-06
Request revised date: 2015-05-27
Online published: 2015-11-10
Copyright
全球范围时空连续的长时间序列地表反照率,对气候模拟与陆面过程研究具有重要意义。针对现有地表反照率产品普遍存在大量的数据缺失、有效反演比例低和时间序列短的问题,本文以多年MODIS和AVHRR数据,通过构建背景知识库进行高时间分辨率的AVHRR和MODIS数据的BRDF参数反演,实现MODIS与AVHRR数据在像元尺度上的定量融合,生成了全球时空连续长时间序列的地表反照率产品。首先,通过假设不同年份同一时期的地表状态不变,利用多年同一时期的MODIS和AVHRR观测数据构造多角度方向反射率,基于BRDF模型反演得到窄波段反照率;然后,通过宽波-窄波转换,得到MODIS的宽波段反照率;最后,结合AVHRR长时间序列优势及MODIS数据多光谱的特点,对二者进行定量融合,生成具有高度一致性长时间序列地表反照率产品。验证结果表明,本文地表反照率产品在地表异质性较小时与SURFRAD地面实测反照率之间具有非常好的一致性,在无积雪覆盖时与MODIS反照率产品之间吻合良好。本文的地表反照率产品无时空缺失,且时间覆盖率得到了极大的提高,能支持气候模式模拟与陆面过程模型进行近30 a来的地气系统模拟研究。
商荣 , 刘荣高 , 刘洋 . 基于背景知识的全球长时间序列反照率反演[J]. 地球信息科学学报, 2015 , 17(11) : 1313 -1322 . DOI: 10.3724/SP.J.1047.2015.01313
Global continuous long-term surface albedo products are of great importance to land surface process and climate modeling research. Problems such as severe data missing and low effective retrieval percentage in the current albedo products made them difficult to meet the requirements of climate modeling perfectly. To solve those problems, this paper made two improvements on MODIS albedo product algorithm. One improvement was to modify the way of composing and selecting enough directional reflectance, which were used for the generation of background BRDF parameters. We assumed that the surface would change little in the same time period during different years. Therefore, all directional reflectance of multi-years for each time period could be used for the composition and selection, in which at least 7 valid directional reflectance for each pixel were used to conduct the BRDF retrieval. Another improvement was the quantitative data fusion between MODIS and AVHRR data, which helped to expand the temporal coverage of albedo products to 30 years. Validation results showed that this new albedo product kept great consistency with the relatively homogeneous SURFRAD site albedos and was almost the same with MODIS snow-free albedo product. These were no spatial or temporal data missing in this new albedo product and the temporal coverage had been greatly improved. Therefore, this new albedo product would be more suitable for the applications of land surface process and climate modeling research.
Key words: surface albedo; BRDF; MODIS; AVHRR; long time series
Fig. 1 Flow chart of long time series albedo retrieval图1 长时间序列地表反照率产品反演流程图 |
Fig. 2 Results of the surface albedo products in this paper图2 本文地表反照率产品反演结果图 |
Tab. 1 Inversion ratios of MODIS albedo products in 2005表1 2005年MODIS反照率产品的反演比例 |
时间 | 完全反演(%) | 部分反演(%) | 缺失(%) | 时间 | 完全反演(%) | 部分反演(%) | 缺失(%) |
---|---|---|---|---|---|---|---|
第1天 | 1.66 | 59.58 | 38.75 | 第185天 | 77.58 | 22.38 | 0.04 |
第9天 | 1.36 | 68.75 | 29.89 | 第193天 | 75.45 | 24.51 | 0.05 |
第17天 | 2.03 | 72.93 | 25.04 | 第201天 | 85.09 | 14.88 | 0.03 |
第25天 | 9.93 | 72.48 | 17.60 | 第209天 | 84.42 | 15.57 | 0.00 |
第33天 | 12.98 | 68.39 | 18.63 | 第217天 | 77.54 | 22.43 | 0.03 |
第41天 | 11.93 | 58.61 | 29.47 | 第225天 | 79.96 | 20.03 | 0.00 |
第49天 | 17.13 | 61.50 | 21.37 | 第233天 | 87.88 | 12.12 | 0.00 |
第57天 | 21.55 | 63.42 | 15.03 | 第241天 | 87.98 | 12.02 | 0.00 |
第65天 | 18.18 | 55.87 | 25.96 | 第249天 | 90.40 | 9.59 | 0.01 |
第73天 | 14.27 | 70.03 | 15.70 | 第257天 | 91.00 | 9.00 | 0.00 |
第81天 | 52.30 | 46.50 | 1.20 | 第265天 | 86.41 | 13.57 | 0.03 |
第89天 | 84.07 | 15.87 | 0.06 | 第273天 | 85.99 | 13.86 | 0.15 |
第97天 | 88.45 | 11.55 | 0.00 | 第281天 | 81.60 | 18.30 | 0.11 |
第105天 | 73.00 | 26.95 | 0.05 | 第289天 | 81.30 | 18.57 | 0.13 |
第113天 | 75.12 | 24.76 | 0.13 | 第297天 | 67.19 | 32.33 | 0.48 |
第121天 | 68.88 | 31.00 | 0.12 | 第305天 | 38.36 | 55.59 | 6.05 |
第129天 | 39.78 | 53.07 | 7.15 | 第313天 | 30.35 | 57.30 | 12.35 |
第137天 | 55.06 | 42.15 | 2.79 | 第321天 | 9.45 | 51.44 | 39.11 |
第145天 | 36.85 | 57.79 | 5.37 | 第329天 | 0.20 | 47.21 | 52.59 |
第153天 | 46.80 | 52.70 | 0.49 | 第337天 | 1.00 | 58.88 | 40.12 |
第161天 | 78.14 | 21.78 | 0.08 | 第345天 | 2.21 | 65.21 | 32.58 |
第169天 | 84.46 | 15.53 | 0.01 | 第353天 | 0.40 | 48.89 | 50.71 |
第177天 | 75.24 | 24.76 | 0.00 | 第361天 | 0.18 | 35.25 | 64.57 |
Fig. 3 Global applicability analysis of the surface albedo products in this paper图3 本文地表反照率算法适用性分析全球分布图 |
Fig. 4 Statistics of the RMSE based on time series surface albedo图4 时间序列反照率均方根误差的统计图 |
Fig. 5 Comparison of the time series of SURFRAD site albedo and surface albedo based on AVHRR data图5 基于AVHRR数据反演的反照率与SURFRAD地面观测反照率的时间序列对比图 |
Fig.6 Comparison of the time series of SURFRAD site albedo, MCD43 surface albedo and surface albedo based on MODIS data图6 基于MODIS数据反演的反照率、MCD43地表反照率与SURFRAD地面观测反照率的时间序列对比图 |
Tab. 2 Validation accuracies based on SURFRAD site albedo表2 地表反照率地面验证精度 |
反照率 | BON站点 | GWN站点 | FPK站点 | PSU站点 | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|
BIAS | RMSE | BIAS | RMSE | BIAS | RMSE | BIAS | RMSE | ||||
基于AVHRR的反照率 | -0.029 | 0.036 | -0.006 | 0.025 | 0.002 | 0.032 | -0.032 | 0.042 | |||
基于MODIS的反照率 | -0.020 | 0.032 | -0.012 | 0.024 | 0.018 | 0.039 | -0.033 | 0.049 | |||
MCD43地表反照率 | -0.020 | 0.038 | -0.016 | 0.027 | 0.006 | 0.020 | -0.012 | 0.031 |
Fig. 7 Scatterplots between algorithm-based albedo and MODIS albedo (MCD43)图7 本文反照率产品与MODIS反照率产品(MCD43)的散点图 |
Fig. 8 Difference images between algorithm-based albedo and MODIS albedo (MCD43)图8 本文反照率产品与MODIS反照率产品(MCD43)的差值图 |
The authors have declared that no competing interests exist.
[1] |
|
[2] |
|
[3] |
|
[4] |
|
[5] |
|
[6] |
|
[7] |
|
[8] |
|
[9] |
|
[10] |
|
[11] |
|
[12] |
|
[13] |
|
[14] |
|
[15] |
|
[16] |
|
[17] |
|
[18] |
|
[19] |
|
[20] |
|
[21] |
|
[22] |
|
[23] |
|
[24] |
|
[25] |
|
[26] |
|
[27] |
|
[28] |
|
[29] |
|
/
〈 |
|
〉 |